Integrating the Design of Structure, Control, and Signal Processing
It is well-known that the various disciplines that design the individual components of the final system are not coordinated, except in an ad hoc way. This talk presents some steps toward the formal integration of Structure, Control, and Signal Processing designs. To integrate structure and control we employ the tensegrity structural paradigm. To integrate signal processing and control we employ our recent work on Information Architecture, where the precisions and locations of all sensors and actuators are coordinated with the control design, which are all dictated by the closed loop performance requirements, including a cost constraint on the hardware. We assume that sensor or actuator costs are proportional to the precision of the instrument. The design constraints are: i) the cost of all sensors and actuators must be less than a specified budget, $, ii) the control energy must satisfy a specified upper-bound, U, iii) the closed loop performance must satisfy a specified covariance upper-bound, Y, of the output error, iv) adjustable parameters of the structure are coordinated with the joint structure/control design to achieve the required performance bounds, Y. This feasibility problem allows one to solve several different optimization problems, simply by reducing iteratively any one of the levels sets (constraint requirements) until feasibility is lost. For example, by minimizing control energy subject to all other constraints one may find which performance bounds require feedback control, and which do not.
Contact: Mallory Neet email@example.com